seagull said:
There is no difference in A.I. learning to make paintings or a human.
This is the #1 yapping point for all AI enthusiasts, and it makes me want to scream. There is a difference. Of course there is a difference. A toddler can tell the difference.
People learn through a lifetime of experience and daily hard fought effort, and their memories of, say, a painting they've seen is filtered through an idiosyncratic lens of personal experience and unique context. Machines on the other hand instantaneously commit everything they've been fed to photographic memory (which at this point is 6 billion images, or the bulk of all human made art throughout the course of history), and then aggregate it to accommodate simple text prompts, often overfitting images pixel for pixel.
So, no. There is no similarity. I wish people would grow up and find some better argument that can't be dismantled by the tiniest bit of critical thinking and common sense. But I understand. There's no room for logic and common sense when you want what you want regardless.
seagull said:
It is simply copying the style but still depends on user input for it to process the painting.
It requires as much user input as does a google search. I'll give you that. But honestly, you don't even need to be able to spell correctly. You could even mash out a string of gibberish and the machine will still produce for you a perfectly rendered image.
Everyone who uses this stuff GROSSLY overstates their contribution to the process. It's minimal to the point of nonexistent. A monkey could do it. I mean, honestly...
seagull said:
However, this can also become alive in the A.I. script itself. And then it might truely develop its own style.
Stop with the anthropomorphizing. It's embarrassing. These are not living, independently thinking beings. They're machines. All they can do is regurgitate what they are fed. What you are suggesting is a fantasy.
seagull said:
Banning this technology will mean you give the upper hand to the others who do use the technology.
This is the #2 yapping point for all AI enthusiasts: "This is progress. And even if it isn't, it's here now and there's nothing you can do about it. So just adapt or get left behind". I say to hell with that. I don't use it because it's theft, and I know a great many principled artists who feel exactly the same.
Your servile "whatcha gonna do?" attitude is going to age very badly. This is the year that these companies burn to the ground and people start going to prison. The NY Times sued OpenAI last week for copyright infringement, and they will most likely win (because they're absolutely correct and have plenty of receipts). That means that Disney will inevitably follow, and then litigation after litigation until these companies are pounded into dust.
Remember Napster? Give it a google for some much needed historical context.
As I posted above, one week prior it was proven that the datasets used to train virtually every currently available text to image generator contain child pornography. The models utilizing them are still operating, but they are only compounding their criminal culpability and will almost certainly be held to account.
And conversations between the CEO of Midjourney and a number of Midjourney devs were leaked just a few days ago (along with a Google doc containing 5,000 new artist names - including at least one longstanding Nexus member I'm aware of), where they discuss incorporating them into V6 and laundering the data in such a way that it won't be traced.
All of this is to say that there is a reckoning coming. I look forward to saying I told ya so.
gameranx.com
